Hello everybody,
I have a few things to ask you related to the calibration with FEE.
It's a lot of stuff, I don't expect an answer right away, maybe we
could discuss some of these topics during today's meeting.
I'd like to give an update next Wednesday, so there's time to
answer.
1)At which point of the simulation can I found "uncalibrated" data?
I mean, for example, after the readout simulation I can plot the
histogram of the seedhit energy, can I have the same histogram but
with the 4096 dst channels values? I need those because..
2)When we'll have real data, I will have to fit those histograms and
associate the fitted values to the calibration points I'll have set.
At which point of the reconstruction will this be done?
3)Up to now, I've been getting the information from the simulated
data out (in a very inelegant way, I must confess) and analyzed and
fitted it with root. When we will have to calibrate the real data,
will it have to be all within hps-java? I've got no idea of how to
perform fits within the java framework....
now about the calibration points
4)I've been simulating the data using these drivers:
<driver name="CalibrationDriver"
type="org.hps.conditions.deprecated.CalibrationDriver">
<driver name="EcalReadout"
type="org.hps.readout.ecal.FADCEcalReadoutDriver">
<driver name="EcalConverter"
type="org.hps.recon.ecal.EcalRawConverterDriver">
<driver name="EcalClusterer"
type="org.hps.recon.ecal.GTPEcalClusterer">
clusters are needed just to select the FEE, the information I use is
the SeedHit energy. Is this procedure ok to set the calibration
points? i.e. is the energy simulated in this way the closest
possible to the real signal?
5)which data should I use to set the points?
I've been doing my studies on this set here:
/mss/hallb/hps/production/slic/beam-tri/2pt2/egsv3-triv2-g4v1_s2d6-readout20140522_[1-2500].slcio
(i.e. the slic output available on tape at JLAb), but if I get it
right, this has been done using v6 geometry. To confront v6, v7 and
v8, I've reconstructed 50 stdhep files ( about 0.1 second of beam)
using the three geometries to confront them. Unfortunately the
statistics is very low, so I can confront only some crystals, but in
those with enough population the differences between the three
geometries is low. For example, having around 5000 entries in the
histrograms, the differece between v7 and v3 is around 1% and
between V7 and V8 is around 0.5%. 1% difference meand a difference
of about 15 MeV in energy, which is pretty low, and I think we can
live with that. What do you think?
6)Once I have the point, where should I put them?
and now on positrons
7)I've discussed with John the possibility to calibrate the shadowed
crystals using positrons in a FEE-fashioned way: taking the seed hit
spectrum of the positrons and dividing it by their momenta given by
the SVT, will give a distribution very similar to the one of FEE.
In the picture you can see it for crystal nr 4 (ix=-20, iy=5).
To generate these histograms I've used the reconstrucetd events
available on tape, specifically the "trigger tridents with pileup".
I think a calibration with positrons is possible, but is way more
complicated since we need information from both the Ecal and the
SVT, so it has to be performed after the reconstruction.
do we want to do it? is it really doable? How precise can it be (I'm
asking the SVT guys, since it depends mostly on the momentum
resolution)? Which MCdata should we use to set the points?
This is pretty much it.
thanks to everybody, any intel will be highly appreciated!
L.